We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent 1003540 commit a89b932Copy full SHA for a89b932
1 file changed
configs/generator/sst2_hf.yaml
@@ -31,7 +31,7 @@ training:
31
# torch_dtype: "auto" # (optional) lets HF pick fastest dtype
32
33
# ── optimiser & scheduler ─────────────────────────────────────────────────
34
- learning_rate: 5e-5 # good starting LR for GPT‑2 on small corpora
+ learning_rate: 0.00005 # good starting LR for GPT‑2 on small corpora
35
warmup_ratio: 0.1
36
37
# ── misc performance knobs ────────────────────────────────────────────────
0 commit comments